Development of Th17-associated interstitial kidney inflammation in lupus-prone mice lacking the gene encoding Signal Transduction and Activator of Transcription-1 (STAT-1)

Abstract

Type I interferon (IFN-I) signaling is a central pathogenic pathway in Systemic Lupus Erythematosus (SLE), and therapeutics targeting IFN-I signaling are in development. Multiple proteins with overlapping function participate in IFN signaling, but the signaling events downstream of receptor engagement are unclear. We employed highly-multiplexed assays to characterize autoantibody production, cytokine/chemokine profiles, and Signal Transduction and Activators of Transcription (STAT) phosphorylation to investigate the individual roles of IFNAR2, IRF9 and STAT1 in MRL/lpr (lpr) mice. Surprisingly, we found that Stat1^−/−, but not Irf9^−/− or Ifnar2^−/− mice, developed interstitial nephritis characterized by infiltration with RORγT⁺ lymphocytes, macrophages and eosinophils. Despite pronounced interstitial kidney disease and abnormal kidney function, Stat1^−/− mice had decreased proteinuria, glomerulonephritis and autoantibody production. Phospho-specific flow cytometry (phosphoflow) revealed shunting of STAT phosphorylation from STAT1 to STAT3/4. In summary, we describe unique contributions of STAT1 to pathology in different kidney compartments, and provide novel insight into tubulointerstitial nephritis (TIN) a poorly understood complication that predicts end-stage kidney disease in SLE patients.

INTRODUCTION

Systemic Lupus Erythematosus (SLE) is a chronic autoimmune disease characterized by inflammatory destruction of multiple organs, including the skin, vasculature, and kidneys. Patients are treated with glucocorticoids, NSAIDs, cytotoxic agents, hydroxychloroquine, mycophenolate mofetil (MMF) and Rituximab, alone or in combination¹. In the last decade, pathophysiological pathways in SLE have been discovered and targeted. These include therapies targeting B cell activator factor (BAFF) and IFNα (a Type I interferon, IFN-I) ².

Evidence that IFNα plays a crucial role in SLE is extensive. A subset of patients administered IFNα for hepatitis or malignancies develop a SLE-like illness, ranging from development of anti-nuclear antibodies (ANA) to frank autoimmunity meeting SLE diagnostic criteria³. These data show that sustained exposure to high levels of IFNα can induce an SLE-like syndrome. Transcript profiling of bulk peripheral blood mononuclear cells (PBMCs) and immune cell subsets from SLE patients demonstrated the presence of an IFNα-inducible gene expression signature or “IFN biosignature” in over half of SLE patients^4,5. The IFN biosignature has been shown to correlate with Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) and the American College of Rheumatology (ACR) 1997 revised SLE criteria⁶. Using the IFN biosignature, patients can be stratified as IFN-high or IFN-low patients.

The IFN biosignature has been robustly recapitulated in murine models of SLE that serve as powerful tools in elucidating the role of pleotropic proteins like IFNα⁷. However, results from groups studying lupus-prone mice lacking the α- chain of IFN-α/βR (called IFNAR1) are inconsistent, reporting either improved or worsened disease in mice of differing genetic backgrounds^8–10. Our group has previously defined important roles for IFNAR2, Interferon Regulatory Factor 9 (IRF9), and Signal Transduction and Activator of Transcription 1 (STAT1, Supplemental Fig. S1) in the pristane murine model^11,12. We identified defects in Toll-like receptor (TLR)-7/9 specific B cell responses and isotype-switched IgG autoantibodies in Ifnar2^−/−, Irf9^−/− and Stat1^−/− mice. These results support a model where IFN signaling is upstream of TLR-specific B cell responses, and is important for the production of isotype-switched IgG autoantibodies.

Because the pristane model is an inducible model associated with mild kidney pathology, the role of IFNAR2, IRF9, and STAT1 in spontaneous development of organ-specific involvement such as glomerulonephritis (GN) is unclear. In the lpr model, SLE-like pathogenesis results from a loss-of-function mutation in the Fas gene (lymphoproliferation, lpr), which leads to defects in negative selection of autoreactive B and T lymphocytes. Lpr mice recapitulate many human SLE pathologies – particularly progressive immune-complex mediated GN seen in lupus nephritis, a leading cause of mortality^13,14. Other features include female bias; diffuse organ inflammation; and autoantibodies directed against DNA and RNA containing antigens¹⁵.

Given the plethora of cytokines that are dysregulated in the promotion of autoimmunity^16–18, growing interest has centered on shared signaling pathways including phosphorylation of STAT family members. Cytosolic STATs are activated through receptor ligation and subsequent tyrosine phosphorylation by Janus family tyrosine kinases (JAKs). This results in translocation of STATs to the nucleus and initiation of transcription of many genes, including those vital to immune function¹⁹.

Here we investigate the individual contributions of the IFN-I/II signaling components IFNAR2, IRF9 and STAT1 to spontaneous SLE pathogenesis in the lpr mouse model. Employing high-throughput methods for characterizing autoantibody profiles and JAK-STAT signaling pathways, we investigate how disease manifestations are impacted by loss of the genes encoding each individual protein. Unexpectedly, we discovered that STAT1 plays a central role in the development of interstitial nephritis mimicking tubulointerstitial nephritis (TIN), which correlates with end-stage renal failure in SLE patients²⁰. We further demonstrate marked infiltration of the kidney interstitium with Th17 lymphocytes; enrichment of Th17 cells in secondary lymphoid organs; alterations in autoantibody profiles; elevated levels of cytokines, chemokines and growth factors; and shunting of STAT1 signaling to other cytokine pathways. Taken together, our results suggest that pharmacological inhibition of STAT1 leads to ameliorated glomerular disease, but may cause tubulointerstitial pathology and other unanticipated effects on the immune system.

RESULTS

IFN-I/II knockout (KO) mice exhibit decreased proteinuria while Stat1^−/− mice have elevated renal function markers

A hallmark of human and murine SLE is the development of GN, in which immune complex deposition in glomeruli results in inflammation and failure to retain proteins^21,22. We hypothesized that mice with defects in IFN signaling would have less severe inflammation and thus would be protected from GN development. We assessed proteinuria in Ifnar2^−/−, Irf9^−/−, Stat1^−/−, lpr and MRL/mpj (mpj) female mice at 16 weeks to allow for disease development in organ systems including the kidneys^23–26. Mpj mice lack the lpr mutation and served as a control strain. As expected, lpr mice developed significantly higher levels of proteinuria compared to mpj controls (p<0.001, Fig. 1a). Mice lacking IFNAR2, IRF9 or STAT1 all exhibited significantly decreased proteinuria compared to lpr mice (p<0.05; p<0.0001; p<0.0001, respectively), indicating relatively preserved glomerular function, consistent with our initial hypothesis.

Fig. 1.

Stat1^−/− mice exhibit decreased proteinuria but abnormal renal function. a: Urine protein levels were determined by Bayer dipsticks in 16-week old mice, Ifnar2^−/−, n= 11; Irf9^−/−, n=30; Stat1^−/−, n=11; lpr, n=27; mpj, n=8. Serum creatinine (b) and Blood Urea Nitrogen (BUN) (c) were measured on a clinical chemistry analyzer. Bars represent mean values and error bars represent the standard error of the mean (SEM). P values were determined by Mann-Whitney test, Ifnar2^−/−, n= 7; Irf9^−/−, n=8; Stat1^−/−, n=7; lpr, n=8; mpj, n=4.

IFNAR2 and IRF9 ablation led to decreased serum levels of both creatinine and blood urea nitrogen (BUN) as compared to lpr mice, although not statistically significant (Fig. 1b–c). As expected, lpr mice had significantly higher serum levels of both creatinine and blood urea nitrogen (BUN) as compared to mpj mice (p<0.01). Surprisingly, Stat1^−/− mice also exhibited increased creatinine and BUN levels as compared to mpj mice, suggesting poor renal function despite improved proteinuria (p=0.01 and p=0.03).

Stat1^−/− mice develop unexpected interstitial kidney disease

Kidney disease in SLE patients can be categorized into distinctive classes in which different areas of the kidney are affected. Because proteinuria is largely a measure of glomerular damage, further studies were performed to assess global kidney morphology. We performed Periodic Acid-Schiff (PAS) staining on sections prepared from paraffin-embedded kidneys (Fig. 2a) and quantified renal nephritis using the NIH Activity and Chronicity Indices in a blinded manner²⁷ (Supplemental Table S1).

Fig. 2.

Stat1^−/− mice display increased interstitial acute and chronic kidney pathology with infiltrating CD45⁺ immune cells. a: Left, 5 µm-thick sections were stained with period acid-Schiff (PAS) and imaged at 20X magnification. Middle, PAS staining imaged at 40X magnification. Right, 5 µm-thick sections were stained for CD45 by immunohistochemistry using 3’Diaminobenzidine (DAB). b–e: Kidneys were scored using the NIH activity and chronicity scoring system (Supplemental Table T1). All knockout mice exhibit decreased active (b) chronic (c) pathology. Stat1^−/− mice exhibit increased acute (d) chronic (e) interstitial pathology. Ifnar2^−/−, n= 7; Irf9^−/−, n=10; Stat1^−/−, n=8; lpr, n=11; mpj=5. Bars represent mean values and error bars represent the standard error of the mean (SEM). P values were determined by Mann-Whitney test.

As we predicted, Irf9^−/− and Stat1^−/− mice exhibited lower acute and chronic glomeruli scores compared to lpr mice, although this result was not statistically significant (Fig. 2b–e). Conversely, Ifnar2^−/− kidneys had increased acute and chronic glomerular scores as compared to lpr kidneys, although this also was not statistically significant. Unexpectedly, Stat1^−/− kidneys showed significantly increased acute and chronic tubulointerstitial pathology compared to lpr kidneys (p=0.0007 and p=0.0002, respectively, Fig. 2d–e), which was not observed in Ifnar2^−/−, Irf9^−/−, and mpj kidneys. To further investigate this finding, we performed immuno-histochemical stains on kidney sections for the leukocyte marker CD45 (Fig. 2a). We observed marked interstitial infiltration of CD45⁺ leukocytes in kidneys from Stat1^−/− mice as compared to Ifnar2^−/−, Irf9^−/−, lpr and mpj mice, suggesting an immune cell-mediated process was responsible for the tubolointerstitial disease.

RORγT⁺ cells infiltrate the interstitium of Stat1^−/− mice

Glomerulonephritis observed in lpr mice is a B cell-associated process characterized by deposition of IgM, IgG, and complement component C3 (C3)^21,22. Recently, elevated levels of Th17 lymphocytes – a subset of inflammatory T cells – has also been described in both the peripheral blood and kidneys of lupus nephritis patients^28–30. To better characterize the immune cell infiltrate within the interstitium of KO mice, we performed a series of immunohistochemical stains. In addition to stains characterizing GN, we performed stains for RORγT (the canonical transcription factor required for Th17 cell development), IBA1 (a macrophage marker) and eosinophils. Both macrohpages³¹ and eosinophils^32–34 have been described in polarization and downstream function of Th17 lymphocytes.

As expected, lpr kidneys showed increased IgM and IgG deposition in the glomeruli as compared to kidneys from BALB/c mice (Fig. 3a). In contrast, minimal staining for IgM and IgG was observed in the glomeruli of Stat1^−/− mice, suggesting that other factors and/or cell types are involved. We observed impressive infiltration of RORγT⁺ cells confined to the interstitium of Stat1^−/− mice (Fig. 3a, right). Stat1^−/− interstitium also demonstrates increased staining for IBA1 (Fig. 3a, right) and eosinophils on hemotaxylin-eosin (H&E) stain (Supplemental Fig. S2). Taken together, histologic evaluation of kidney specimens implicates a role for RORγT⁺ cells in mice lacking STAT1, in contrast with the central role played by B lymphocytes in lpr mice with GN.

Fig. 3.

Mice genetically-deficient for IFN-signaling components display differential accumulation of cells expressing RORγT⁺ in the kidney interstitium and IL-17A/IFNγ in secondary lymphoid organs. a: 7 µm-thick frozen sections were stained for RORγT (red), IBA1 (magenta), IgM (green) and DAPI (blue). Stat1^−/− mice exhibit increased Th17 and macrophage infiltration in the interstitium. b: Intracellular staining of spleen (top panels) or lymph node (bottom panels) for IFNγ (x-axis) and IL-17A (y-axis) show increased frequencies of IL-17A producing CD4⁺ T cells in Stat1^−/− mice and increased frequencies of IFNγ-producing CD4⁺ T cells in Irf9^−/− and Ifnar2^−/− mice. Data for BALB/c mice with and without STAT1 are shown in Supplemental Fig. S3. c/d: Graphical representation of flow cytometry data from spleen. Ifnar2^−/−, n=7; Irf9^−/−, n=7; Stat1^−/−, n=7; lpr, n=8. P values were determined by Mann-Whitney test. Error bars represent SEM.

Th1/Th17 polarization is altered in IFN-I/II signaling KO mice

In addition to their role in kidney disease, the circulating Th17 cell population is increased in SLE patients³⁵. We hypothesized that ablation of STAT1 would lead to increased Th17 polarization in secondary lymphoid organs. To test this hypothesis, we quantified the proportions of IL-17A and IFNγ producing lymphocytes in the spleen and lymph nodes (axillary, mandibular and mesenteric) of Ifnar2^−/−, Irf9^−/−, Stat1^−/−, and lpr mice by flow cytometry.

The proportion of IL-17A⁺ CD4⁺ T cells was increased in Stat1^−/− splenocytes by more than eight fold compared to lpr splenocytes (0.84% vs 0.10%, respectively) (Fig. 3b) and was significantly increased across multiple mice (p=0.007, Fig. 3c). No significant changes were observed in the frequencies of IL-17A producing CD4⁺ T cells in Ifnar2^−/− and Irf9^−/− (0.15% and 0.04%, respectively) as compared to lpr mice (Fig. 3b). To test whether absence of STAT1 alone would lead to expansion of the Th17 lymphocyte population, intracellular cytokine staining (ICS) of IL-17A was also performed on BALB/c splenocytes lacking STAT1. No significant enrichment was found when compared to wildtype BALB/c mice (Supplemental Fig. S3).

As expected, fewer IFNγ-producing CD4⁺ T cells were observed in spleen and lymph nodes obtained from Stat1^−/− mice as compared to cells obtained from lpr mice (12.12% and 30.02% from spleen, respectively) (Fig. 3b) and was significantly decreased across multiple mice (Fig. 3d, p=0.004). Ifnar2^−/− and Irf9^−/− mice had significantly larger populations of IFNγ-producing CD4⁺ T cells in both spleens and lymph nodes (42.56% and 53.6% from spleen, respectively) (Fig. 3b), also significant across multiple mice (p=0.0185, Fig. 3d). These results suggest that crosstalk between IFN-I and IFN-II signaling occurs where in the absence of IFN-I signaling proteins, a compensatory increase in IFNγ signaling is observed, likely mediated by STAT1 which is shunted from IFN-I to IFN-II signaling.

Stat1^−/− mice exhibit decreased IgM/G autoantibody production against SLE-associated antigens

Antibodies directed against autoantigens are clinically relevant in GN³⁶, while the role of autoantibodies in interstitial kidney disease is not well understood. We employed protein microarrays to profile autoantibodies in the sera of Stat1^−/−, Ifnar2^−/−, Irf9^−/−, and lpr mice. This method has been well-established by our lab for studying human^37–42 and murine SLE^11,12,43. In brief, arrays containing over 600 features and comprised of more than 100 unique proteins and peptides that are known or putative SLE autoantigens were printed, probed with sera derived from knockout mice and bound autoantibodies were detected with a fluorophore-conjugated secondary antibody. Arrays were scanned, and fluorescence intensity was quantitated. The Significance Analysis of Microarrays (SAM) algorithm⁴⁴ was then employed to determine antigens with statistically significant differences between groups of mice, and a hierarchical clustering program grouped individual mice based on similar autoantibody profiles.

IgM autoantibody reactivity against SLE-associated autoantigens was decreased in Stat1^−/− as compared to lpr mice (Fig. 4a). SAM identified 12 of these IgM autoantibodies as significantly decreased in Stat1^−/− sera compared to lpr sera. Differentially-targeted autoantigens included U1-snRNP, histones, dsDNA, CENP-B, and PM/Scl75, many of which are important in human SLE. SAM identified a similar but constricted subset of IgG autoantibodies that were significantly decreased in Stat1^−/− compared to lpr sera (Fig. 4b).

Fig. 4.

Stat1^−/− mice exhibit decreased autoantibody production against SLE-associated antigens. Individual autoantigen arrays containing over 600 features and 100 antigens were incubated with serum obtained from lpr, Ifnar2^−/−, Stat1^−/−, and Irf9^−/− mice. The SAM algorithm was used to determine antigen features with statistically significant differences in reactivity between sera derived from either lpr, Stat1^−/−, Irf9^−/− or Ifnar2^−/− mice. Hierarchical clustering of samples based on reactivity to antigens with statistically significant differences is displayed as a heatmap and dendrogram. Sera from Stat1^−/− and lpr mice were used to probe autoantigen arrays and detected with anti-IgM (Fig. 4a) and anti-IgG secondary (Fig. 4b) antibodies. The same methods employed in Fig. 4a and 4b were employed with sera from Irf9^−/− and lpr mice (Supplementary Figure S3). SAM did not identify any features with statistically significant differences between Ifnar2^−/− and lpr mice. c: Confirmation of SAM-identified IgG (Fig. 4c) and IgM (Supplemental Fig. S5) autoantibodies performed by ELISA. Data represent time resolved fluorescence for serum samples obtained from individual mice. Horizontal bars represent mean counts for each group. P values were determined by Mann-Whitney test. lpr, n=5, Stat1^−/−, n=5. d/e: Stat1^−/− or lpr B cells were stimulated with or without IFNα for 4 hours and relative expression of Mx1, TLR7 and TLR9 was measured. P values were determined by Mann-Whitney test.

When comparing Irf9^−/− and lpr sera, SAM identified increased IgM and IgG autoantibody reactivity against histones in Irf9^−/− sera (Supplemental Fig. S4a and S4b). These data reveal a potentially unique role for IRF9 in antibody production against histones. We did not identify any antigens with significantly different reactivity when comparing Ifnar2^−/− and lpr sera. All significant array findings were validated using enzyme-linked immunosorbent assay (ELISA). Of 22 SAM-identified antigens identified in the entire study across all strains, 17 were confirmed by ELISA to have significant differences in titer between KO and lpr mice (Fig. 4c, and Supplemental Figs. S4c and S5).

STAT1 is required for the expression and upregulation of TLR7 by IFNα

Nucleic acid components of SLE autoantigens can activate autoreactive B cells via TLR7, which recognizes RNA, and TLR9, which recognizes DNA. TLR expression is also upregulated by IFN-I secreted by pDCs in human naive B cells and modulates isotype switching⁴⁵. We have previously demonstrated in the pristane model that mice lacking IFNAR2, IRF9 or STAT1 had defects in upregulating transcripts encoding TLR7 and/or TLR9 in response to IFNα exposure or TLR agonists^11,12. Moreover, Stat1^−/− pristane mice had a marked defect in isotype switching of autoantigen-specific B cells, but not B cells immunized with ovalbumin. We hypothesized that the decreased autoantibody production observed in Stat1^−/− lpr mice was due to a similar defect in TLR7/9 upregulation.

Splenic B cells were obtained from Stat1^−/− and lpr mice, cultured in the presence or absence of IFNα, and relative mRNA expression levels of TLR7, TLR9 and Mx1 (a known IFNα-inducible gene) normalized to β2M and GAPDH were determined by realtime quantitative PCR. The expression of TLR7, but not TLR9, was significantly upregulated in IFNα-treated B cells from lpr mice, but this upregulation was almost completely abolished in Stat1^−/− B cells (Fig. 4d, left). Therefore, STAT1 is required for IFNα-inducible expression of TLR7 but not TLR9 in murine B cells. Because Stat1^−/− mice have decreased reactivity against both DNA- and RNA-associated antigens, these data suggest overlap of RNA/DNA-sensing by TLR7/9, perhaps through differential endosomal trafficking in the absence of STAT1⁴⁶. Alternatively, it is possible that defects in autoantibody production in Stat1^−/− mice result from TLR-independent pathways.

As expected, induction of expression of Mx1 mRNA was significantly decreased in B cells from Stat1^−/− mice (p=0.016, Fig. 4e) compared to lpr B cells. However, Stat1^−/− B cells stimulated with IFNα retain the ability to significantly upregulate Mx1 expression as compared to unstimulated cells (p=0.02, Fig. 4e), revealing that STAT1 is dispensable for IFNα signaling. This finding supports work demonstrating high redundancy and plasticity of STAT phosphorylation in different physiological states^47,48. We tested the hypothesis that IFNα signaling can signal through molecules other than STAT1 using phosphoflow cytometry.

Disruption of genes encoding IFN-I/II signaling proteins results in shunt STAT phosphorylation to different cytokine signaling pathways

In addition to the finding that SLE patients have abnormal levels of many serum cytokines and chemokines, many of which are interferon inducible, several publications have demonstrated that dysregulated STAT signaling plays a role in murine models of SLE and in human SLE patients^49,50. We hypothesized that both Th1/Th17 polarization (Fig. 3b-3d) and transcription events downstream of IFNα signaling (Fig. 4d-4e), observed in the absence of IFN-I/II signaling proteins could be attributed to dysregulated STAT signaling.

Splenocytes from individual mice were isolated and stimulated with individual cytokines for 20 minutes. We then used phosphoflow cytometry to measure phosphorylation of STATs 1, 3, 4, 5, and 6 in CD4⁺ T cells (CD3⁺CD4⁺CD8⁻), CD8⁺ T cells (CD3⁺CD4⁻CD8⁺), and B cells (CD3⁻CD4⁻CD19⁺) from Stat1^−/−, Ifnar2^−/−, Irf9^−/−, and lpr mice. We selected cytokines thought to play critical roles in SLE pathogenesis including IFNα, IFNγ, IL-21, and IL-27^51–53. To identify additional cytokines to include in the stimulation panel, we performed a multiplex bead-based assay on Stat1^−/−, Ifnar2^−/−, Irf9^−/−, and lpr serum to determine levels of 26 different cytokines and chemokines (Supplemental Fig. S6). Compared to lpr sera, sera from Stat1^−/− mice had significantly increased levels of TNFα, IL-4 and eotaxin, in addition to significantly decreased levels of IL-12p40, a cytokine that is critical for Th1 polarization^54,55. Irf9^−/− sera also had significantly increased levels of IL-4, as well as IL-1a when compared to lpr sera. Based on these findings, IL-4 and IL-12 were added to our stimulation panel.

Phosphoflow profiling resulted in 1,980 data nodes, where each node represents phosphorylation levels of an individual STAT protein, in response to a single stimulation in a unique cell subset. As expected, Stat1^−/− mice had no STAT1 phosphorylation across all cell subsets (Fig. 5a). In response to IFNα, STAT3 phosphorylation was significantly increased in all subsets of Stat1^−/− mice compared to lpr mice (Supplemental Fig. S7). IFNα stimulation also resulted in significantly increased STAT4 phosphorylation both in CD4⁺ and CD8⁺ T cells of Stat1^−/− compared to lpr mice. These data strongly suggest that IFNα can signal through STAT3/4 and potentially other factors.

Fig. 5.

Targeted deletion of genes encoding IFN-I/II signaling proteins results in shunting of STAT phosphorylation downstream of cytokine stimulation. Splenocytes from Stat1^−/−, Irf9^−/−, Ifnar2^−/−, and lpr mice were stimulated with IFNα, IFNγ, IL-4, IL-21, or IL-27. Phosphorylation of STATs 1, 3, 4, 5, 6 in CD4⁺ T cells, CD8⁺ T cells, or CD19⁺ B cells, was measured by phosphoflow cytometry. a: Fold change of cytokine stimulated over unstimulated condition for individual knockout or lpr mice, shown in heatmap form (data were Log2 transformed). Yellow and blue represent increases or decreases in phosphorylation levels compared to baseline, respectively. Four representative mice were selected per group. b: Phosphorylation of individual STAT proteins downstream of an individual cytokine in one cell subset was compared to corresponding data in lpr mice. Uncorrected p-values determined by Mann-Whitney test are represented in heatmap format. Direction of increased or decreased STAT phosphorylation in comparison to lpr is represented by blue and purple, respectively. Ifnar2^−/−, n=6; Irf9^−/−, n=4; Stat1^−/−, n=6; lpr, n=6. Bar charts of data indicating significance of fold change are presented in Supplemental Fig. S7–S9.

Irf9^−/− mice had increased STAT3 phosphorylation in CD8⁺ T cells as well as in CD19⁺ B cells after IFNα stimulation. Significantly increased STAT4 phosphorylation was observed in CD4⁺ and CD8⁺ T cells from Irf9^−/− mice in response to IFNγ stimulation compared to lpr T cells (Supplemental Fig. S8). This result may explain our finding that Irf9^−/− mice have higher frequencies of IFNγ producing CD4⁺ T cells (Fig. 3b and 3d), as increased signaling in response to IFNγ exposure may be part of a positive feedback loop that drives the development of IFNγ-secreting T cells.

All cell subsets in Ifnar2^−/− mice compared to lpr mice exhibited significantly reduced phosphorylation of STAT1, 3, 4, and 5 in response to IFNα stimulation (Supplemental Fig. S9). These results support a model where the IFNAR2 component of the IFNAR dimeric receptor complex is necessary for receptor signaling. Interestingly, enhanced phosphorylation of STAT5 in response to IL-27 was significantly and consistently observed both in CD4⁺ and CD8⁺ T cells of Ifnar2^−/− mice compared to lpr mice. This result shows that in the absence of IFNAR2, IL-27 signaling can be modulated from phosphorylation of STAT1/3 to phosphorylation of STAT5. It serves as additional evidence of STAT redundancy in different biological states.

DISCUSSION

Using highly multiplexed analyses we show how loss of individual IFN-I/II signaling molecules IFNAR2, IRF9, and STAT1 affects renal disease, autoantibody profiles, T cell subsets, and cytokine signal transduction. Notably, genetic deletion of STAT1 leads to decreased autoantibody production and GN, but induces severe interstitial inflammation accompanied by infiltration of RORγT⁺ cells, CD45⁺ leukocytes, macrophages, and eosinophils in the interstitium. STAT1 ablation results in altered signal transduction characterized by shunting to alternative STAT molecules including STAT3, a key transcription factor involved in Th17 cell differentiation. Together, our findings demonstrate a critical role for STAT1 in the development of Th17 associated autoimmune interstitial kidney disease in a murine model of SLE.

Prevention and treatment of kidney disease is crucial for clinicians who care for SLE patients. Despite a reported prevalence of TIN in SLE kidney biopsies ranging from 23% by Park et al.⁵⁶ up to 72% by Hsieh et al.²⁰, very little is known regarding the immunologic mechanisms underlying TIN. Most histologic measures focus on glomerular involvement to inform therapeutic decisions, but have suboptimal predictive value in renal outcomes. In contrast, multiple groups have shown that SLE patients with TIN are at greatest risk of renal failure as defined by initiation of chronic dialysis treatment or requirement for renal transplant^20,57. Our results lend support to this model as only Stat1^−/− mice with predominant TIN, and not Ifnar2^−/−, Irf9^−/− or lpr mice with predominant GN, showed elevation in kidney markers BUN and creatinine. These results suggest that TIN, more so than GN, contributes to decreased renal function.

IL-17A is thought to play an important role in SLE pathogenesis^58–60. The proportion of Th17 cells in blood of SLE patients is elevated and correlates with SLEDAI and with the presence of lupus nephritis^61,60. IL-17 levels are also elevated in SLE sera⁶². Organ-specific expression of IL-17A-producing cells, particularly in the kidney, is more controversial. Wang et al. employed single cell laser capture microdissection to study cytokine expression levels in glomeruli versus interstitium and observed increased IL-17A mRNA expression both in cells isolated from glomeruli and interstitium of SLE patients. They were unable to detect IL-17A in kidneys by immune-histochemical staining in contrast to Crispin et al. who successfully detected IL-17A by similar staining methods⁶³.

Because staining for IL-17A in murine systems has been difficult and results are controversial, we instead focused on the transcription factor RORγT, a canonical regulator of Th17 cell differentiation. We demonstrated RORγT staining in renal interstitium with concurrent staining of IBA1 in Stat1^−/− but not in lpr mice. FACS analysis demonstrated significantly higher proportions of Th17 cells in lymph nodes and splenocytes derived from Stat1^−/− mice as compared to lpr mice. These data robustly support a role for STAT1 in inhibiting Th17 lymphocyte polarization and trafficking to kidney interstitium in lpr mice.

Costaining of eosinophils with RORγT⁺ cells in the interstitium of Stat1^−/− as compared to lpr mice further supports Th17 lymphocyte activity as IL-17A upregulates various cytokines that promote eosinophillic inflammation^32–34. Multiplexed cytokine profiling studies also revealed increased levels of serum eotaxin, an eosinophil chemotactic chemokine, in Stat1^−/− as compared to lpr mice. Taken together, these data support our hypothesis that interferon signaling plays a role in Th17 cell differentiation and SLE pathology, and that STAT1 in particular plays a central role in Th17-associated TIN.

Despite our findings, our study is limited by the inability to dissect potential independent effects of STAT1 ablation on immune and renal parenchyma cells. Previous work has described elevated levels of IFNγ mRNA in lpr renal cells themselves^64,65, resulting in the increased production of inflammatory cytokines like CSF-1 and TNF-α which can induce apoptosis of tubular epithelial cells involved in interstitial homeostasis⁶⁶. Our finding of severe Th17-associated tubulointerstitial disease is unexpected in the setting of STAT1 ablation since IFNγ signaling requires STAT1 homodimerization and translocation to the nucleus where it binds to gamma activated site (GAS)-containing genes. Adoptive transfer of Stat1^−/− hematopoietic stem cells (HSCs) or IL-17 producing cells into lpr mice was unsuccessful due to difficulty with breeding and radiosensitivity of the lpr background strain. Further experiments are planned to shed light on the independent role of STAT1 in renal versus hematopoietic cells.

It should be noted that disease severity in our mouse colony was observed to be less pronounced than other published studies, but does not affect the conclusions we have reached. Several possibilities exist for this discrepancy. Variability in disease development within strains housed in different facilities has been reported. Factors such as diet, sex of handler, and site of blood collection (retro-orbital, vena cava or tail vein) can impact both behavioral and physiologic readouts^{55–57,67–69}. Finally, nephritis development is dependent on age at analysis. Previous studies have euthanized mice at 20 – 24 weeks where disease severity and mortality is high. In accordance with guidelines of the Institutional Animal Care and Use Committee of Stanford University, euthanasia of 16-week-old mice was recommended as it allowed for both full development of disease manifestations and humane care.

Analysis of cytokine production and STAT phosphorylation in Stat1^−/− splenocytes allowed us to investigate signaling mechanisms that could in part help to explain differences observed in T-helper cell polarization. We hypothesized that Th17 cell enrichment was associated with compensatory STAT3/4 phosphorylation, both previously described in Th17 cell polarization^70–72. Our phosphoflow profiling data support this hypothesis as Stat1^−/− splenocytes exhibited shunted phosphorylation to STAT3/4 downstream of multiple stimuli including IFNα, IL-21 and IL-27, and across multiple cell subsets. Phosphorylation of other STAT proteins in response to IFNα supports our qPCR data showing that Stat1^−/− B cells retain the capacity to upregulate the transcription of the interferon-inducible gene Mx1 upon IFNα receptor ligation, presumably mediated by STAT3/4.

Phosphoflow analysis of cells from Ifnar2^−/− mice shows little if any STAT phosphorylation upon IFNα stimulation. Despite lack of IFNα signaling, Ifnar2^−/− mice still have prominent autoantibody production and moreover have marginally increased kidney pathology, both of which have previously been associated with high IFN signaling rather than absence of IFN signaling^6,7. One potential explanation for the discrepancy between phenotypes of Ifnar2^−/− and Stat1^−/− mice is that lpr pathology is driven by IFNγ, which requires STAT1 for signal transduction through its receptor, and that IFNα is less influential⁶⁵. However, evidence shows that both IFNα and IFNγ can induce the formation of the IFN-stimulated gene factor 3 (ISGF3) complex and bind to IFN-stimulated Response Element (ISRE) and gamma activated site (GAS)-containing genes, repectively^73,74. IFNα- and IFNγ-inducible genes have ~25% overlap, and genes most highly upregulated by IFNα are also IFNγ-inducible⁷⁵.

Differences exist between the biology of IFN-I and IFN-II signaling proteins. IFNAR1/2 have both been shown to either directly or indirectly interact with other proteins including TAM family members, which are thought to inhibit IFN-I-induced inflammation⁷⁶. It is possible that loss of IFNAR2 inhibits both anti-inflammatory TAM signaling in addition to inflammatory IFN signaling, resulting in a net-neutral effect. Treatment of lpr mice with a monoclonal anti-IFNAR antibody ameliorates disease at early time points. At later stages of disease, IFN-I signaling-independent pathways overcome the therapeutic effects⁷⁷. Mangini et al posit that one of these pathways is downstream of IFNγ, suggesting important interactions between IFN-I/II signaling pathways. We demonstrate direct evidence of crosstalk with increased IFNγ production in the secondary lymphoid organs of Ifnar2^−/− and Irf9^−/− mice. Crosstalk is likely attributed to shared phosphorylation of STAT1 downstream of IFNα and IFNγ in addition to promiscuity of STAT phosphorylation.

Our studies highlight the complexity of IFN-I/II signaling and the necessity for mechanistic studies in implementing therapies targeting IFNs and JAK-STATs. Our Stat1^−/− model of Th17-associated TIN may lead to better understanding of JAK-STAT inhibitors like Tofacitinib, which inhibits JAK1/3⁷⁸. While there are no reports of Tofacitinib-induced TIN, no murine studies have addressed the effects of JAK1/3 inhibition on autoimmune renal diseases. IFN-targeted therapies Sifalimumab (MEDI-545) and Rontalizumab (RG7415) are still in early-stage clinical trials, and suggest that IFN inhibitors may be beneficial in certain subsets of patients^79–81. Our results underscore the potential benefits in subsetting disease not only on IFNα signaling but also signaling of other cytokines like IFNγ and IL-21. Future human studies will benefit from proteomic analysis to understand global cytokine signaling in patients to guide trial design and to characterize mechanistic effects of targeted therapies. Taken together, the present study reveals unique contributions of IFN signaling proteins to SLE pathogenesis and highlights a role for STAT1 in interstitial kidney disease in SLE. These results have the potential to improve our understanding of clinically important TIN and to guide studies of targeted cytokine therapies in patients.

METHODS

Mice

Ifnar2^−/− mice on the BALB/c background were kindly provided by Paul J Hertzog (Monash University, Clayton, Australia). Irf9^−/− mice on the BALB/c background were purchased from RIKEN BioResearch Center. Stat1^−/− mice on the BALB/c background were a generous gift from J. Durbin (Ohio State University, Columbus, Ohio, USA). All knockout mice were backcrossed onto the lpr (Jackson Laboratories) background to the N8-N10 generation. Mice used in this study were maintained under standard conditions at the Stanford University Research Animal Facility. Data presented are from female mice euthanized at 16 weeks of age. Due to the characteristically small litter sizes and poor breeding behavior of lpr mice, the use of littermate wild type controls was not always possible. When breeding did not yield sufficient numbers of wild-type controls, age- and sex-matched lpr mice were substituted. All unique repeat experiments included both knockout and wild type mice. To ensure accuracy in breeding, every mouse analyzed in this study was genotyped for either homozygous ablation or presence of the desired gene (Supplemental Table T2). Proteinuria was monitored by dipstick analysis using Albustix (Bayer). Serum creatinine and BUN analysis was performed on Siemens Dimension Xpand analyzer. All animal experiments were approved by, and performed in compliance with, the guidelines of the Institutional Animal Care and Use Committee of Stanford University.

Histopathology and immunohistochemical staining

Mice were euthanized by CO₂ exposure followed by cervical dislocation. Kidneys were surgically removed, fixed in buffered formalin, embedded in paraffin, sectioned, and stained with either periodic acid-Schiff (PAS), Hemotaxylin and Eosin (H&E) or CD45 (30-F11, BD Pharmingen; performed by Histotec, Hayward, California, USA). The slides were scored in a blinded fashion by one of the authors (JF) according to NIH kidney scoring (Supplementary Table 1), both for acute and chronic disease indices.

Lupus autoantigen microarrays

Detailed autoantibody profiling protocols and a list of arrayed antigens have been previously published³⁹. Briefly, SLE-associated autoantigens were printed at 0.2 mg/ml in ordered arrays on nitrocellulose-coated FAST slides (Whatman, Piscataway, New Jersey) using a VersArray ChipWriter Pro Robotic Arrayer (Bio-Rad). Individual arrays were blocked with PBS containing 3% FCS and 0.05% Tween 20 (Sigma-Aldrich) for 1.5 hrs on a rocking platform at room temperature. Arrays were probed with 400 µl mouse serum diluted 1:250 in 1X PBST with 5% FCS for 1.5 hrs on a rocking platform at 4°C, followed by washing and incubation with a 1:2000 dilution of cyanine-3-conjugated goat anti-mouse IgM or IgG secondary antibody (Jackson ImmunoResearch Laboratories). Arrays were scanned using a GenePix 4000B scanner (Molecular Devices) at constant PMT power for all arrays. The net mean pixel intensities of each feature were determined using GenePix Pro 6.1 software (Molecular Devices, Sunnyvale, California). Array data will be uploaded to the GEO database upon publication of the manuscript.

Enzyme-linked immunosorbent assays

96 well plates (NUNC MaxiSorp) were incubated at 4°C overnight with 100 µL/well protein solution at 2µg/ml (U1, His1000, H1, H2A, H2B, H3, dsDNA plasmid, dsDNA salmon sperm, dsDNA calf thymus, Intrinsic factor, MPO, Scl70, and CENPB). Plates were then washed 3X with phosphate buffered saline with 0.05% Tween 20 (PBST) and blocked with 200µL/well PBST with 5% FCS for 2 hrs at room temperature. After washing, plates were incubated with 100 µL/well of sample at three dilutions (1:400; 1:800; 1:1600) at 4°C overnight, followed by washing and development using Europium labeled anti-mouse IgG and DELFIA Enhancement solution (PerkinElmer). For IgM autoantibody measurement, endogenous biotin was blocked in two steps by first incubating with streptavidin followed by biotin to block free binding sites on plate bound streptavidin. Samples were then incubated with DSB-X biotin goat anti-mouse IgM (Invitrogen), followed by washing with Europium labeled streptavidin, then developed using DELFIA enhancement solution. Stat1^−/−, n=5; Irf9^−/−, n=5, lpr, n=5, BALB/c, n=2. Detection limit was calculated as 2X the standard deviation of the blanks.

Real-time quantitative PCR

Splenocytes were harvested from Irf9^−/− and Stat1^−/−, and lpr mice. B cells were negatively selected using magnetic beads (Miltenyi Biotech). Cells were more than 95% pure, as assessed by flow cytometry (CD19⁺IgM⁺biotin⁻). B cells were cultured in RPMI supplemented with L-glutamine (2 mM), sodium pyruvate (1 mM), nonessential amino acids (0.1 mM), penicillin (100 U/ml), streptomycin (0.1 mg/ml), 2-ME (5 × 10⁻⁵ M), and FBS (10%) in the presence or absence of 1,000 IU/ml recombinant IFN-α (Calbiochem) for 4 hours. RNA was extracted using RNeasy Mini kit (Qiagen). RNA (10 ng) was amplified using 1-step QuantiTect SYBR Green RT-PCR (Qiagen) and 0.5 µM forward and reverse primers using an Opticon2 continuous fluorescence detector (MJ Research). The fold change in expression of each transcript normalized to β2M and GAPDH was determined by the 2^−ΔΔCt method. QuantiTect Primer Assay sets for murine TLR7, TLR9, MX1, B2M and GAPDH were purchased from Qiagen.

Immunofluorescence staining

Animals were injected intraperitoneally with ketamine (100 mg/kg) and diazepam (5 mg/kg) and monitored. Upon loss of nociceptive reflexes, animals were transcardially perfused with 30 mls of PBS/EDTA followed by 30 mls of 4% paraformaldehyde (wt/vol) in 0.1 M PBS at room temperature. Kidneys were removed and post-fixed in 4% paraformaldehyde at 4°C for up to 24 hrs then cryoprotected in 24% sucrose solution (wt/vol) in PBS for 24 hrs. Kidneys were embedded in Optimal Cutting Temperature compound (Tissue-Tek) and frozen at −80°C, then cut at 7-µm-sections. Sections were allowed to thaw at 20–24°C and rehydrated in PBS for 2 hrs then treated with 0.3% Triton in PBS for 30 mins. 300 ul of ice cold acetone was added to each section and place at −20 for 15 minutes. Immediately, wash with PBS three times. Blocking was performed with 5–10 µg ml⁻¹ purified antibody to CD16/32 for 2 hrs. Primary antibody staining was performed overnight at 4°C for the following molecules: RORγT (AFKJS-9, eBioscience), Iba-1 (cat. #019–19741, Wako) and IgM (1020-01, SouthernBiotech). Secondary antibody staining was performed using donkey anti-goat Alexa Flour 647, donkey anti-rabbit Alexa Flour 568 or donkey anti-rat Alexa Flour 488 (Molecular Probes) in the dark for 1 hr at room temperature. All sections were analyzed by confocal microscopy using either a Leica SP2 AOBS Confocal Microscope or Leica SP8 Confocal Microscope. All images presented are maximum intensity projections of z stacks of individual optical sections.

Intracellular cytokine staining

Splenocytes and lymph nodes (axillary, submandibular, and mesenteric) were harvested in 4°C-chilled MACs buffer (PBS containing 1% FCS and 1 mM EDTA). Using the back of a sterile 5 ml syringe, tissues were manually disassociated in a culture plate containing 10 mls of MACs buffer. Cell suspension was filtered through a 70 µM filter with an additional 20 mls of MACs buffer and pelleted at 500g for 5 mins at 4°C. The pellet was treated with 5 mls of RBC lysis buffer (cat. # 00–4333-57, eBioscience) for 5 mins at room temperature. Lysis was quenched with 30 mls of MACs buffer and the cell suspension was filtered through a 70 µM filter and a 40 µM filter, then pelleted using the above conditions. Cell count was determined using a Z2 Beckmann Coulter counter. Cells were cultured at 1 × 10⁶ cells/ml in RPMI supplemented with _L-glutamine (2 mM), sodium pyruvate (1 mM), penicillin (100 U/ml), streptomycin (0.1 mg/ml), and FCS (10%). Cells were stimulated for 4 hrs with PMA (phorbol 12-myristate 13-acetate; 50 ng/ml), ionomycin (Sigma, 750 ng/ml), brefeldin-A (Sigma, 20 ug/ml), then stained with antibodies against surface markers CD3ε, CD4, CD8a (eBioscience) and Aqua Amine Live/Dead (Invitrogen) for 30 mins. After washing, cells were fixed and permeabilized (cat. #00–8333-56, ebioscience) for 20 mins in the dark and stained with fluorophore-conjugated antibodies against intracellular IL-17A and IFNγ (eBioscience). Following washing, cells were fixed with 1.6% paraformaldehyde and analyzed using an LSR II (Becton Dickinson).

Luminex multiple cytokine/chemokine assay

This assay was performed in the Human Immune Monitoring Center at Stanford University. Mouse 26-plex kits were purchased from Affymetrix and used according to the manufacturer’s recommendations with modifications as described below. Briefly, samples were mixed with antibody-linked polystyrene beads on 96-well filter-bottom plates and incubated at room temperature for 2 hr followed by overnight incubation at 4°C. Room temperature incubation steps were performed on an orbital shaker at 500–600 rpm. Plates were vacuum filtered and washed twice with wash buffer, then incubated with biotinylated detection antibody for 2 hr at room temperature. Samples were then filtered and washed twice as above and resuspended in streptavidin-PE. After incubation for 40 minutes at room temperature, two additional vacuum washes were performed, and the samples were resuspended in Reading Buffer. Each sample was measured in duplicate. Plates were read using a Luminex 200 instrument with a lower bound of 100 beads per sample per cytokine.

Cytokine stimulation and phosphoflow cytometry

Splenocytes were cultured for at least 2 hours and stained for viability with Aqua Amine Live/Dead (Invitrogen) for 30 mins. Following washing and resting for 30 minutes, splenocytes were stimulated with IFNα (PBL Interferon Source, 1000 U/ml), IFNγ (Peprotech, 50 ng/ml), IL-4 (Peprotech, 50 ng/ml), IL-12 (Peprotech, 50 ng/ml), IL-21 (Peprotech, 50 ng/ml), or IL-27 (R&E Systems, 50 ng/ml) for 20 minutes at 37°C. Stimulation was halted with 1.6% paraformaldehyde for 10 minutes at room temperature. Following washing with FACS buffer (0.5% Bovine Serum Albumin and 0.09% Sodium Azide in PBS), splenocytes were permeabilized using pre-chilled MeOH for 20 minutes at 4°C. Following washing, cells were stained for 1 hour at room temperature with fluorophore-conjugated surface antibodies: CD3 (17A2) eFuor450, CD4 (RM4–5) Alexa 700, CD8a (53–6.7) PerCP-Cy5.5, CD19 (1D3) PE-Cy7, NKp46 (29A1.4) APC (all surface antibodies were purchased from eBioscience except NKp46, which was purchased from BioLegend) in addition to intracellular antibodies against phospho-proteins Stat1 Y701 (4A) Alexa 488, Stat3 Y705 (4) Alexa 647, Stat4 pY694 (38) PE, Stat5 Y694 (47) Alexa 488, and Stat6 Y641 (J71–773.58.11) Alexa 647 (BD Biosciences). Following washing, cells were fixed with 1.6% paraformaldehyde and analyzed using an LSR II (Becton Dickinson).

Statistical methods

Microarray data were expressed as mean net fluorescence intensity (MFI) units, representing the mean values from six replicate antigen features on each array. Non-reactive samples were defined as having a maximum normalized IgM or IgG MFI of less than 1,000 for a given antigen. The same methods were applied to Luminex multiple bead based assays. For both, Significance Analysis of Microarrays (SAM)⁴⁴ was applied to the dataset (with the MFI value of undetected array features set to 1) using the Wilcoxon signed-rank test statistic to identify antigens or cytokines/chemokines with statistically significant differences in array reactivity between different groups of mice at FDR of 0 (q<0.001). Binding reactivity heatmaps were generated using MultiExperiment Viewer (MEV TM4 Microarray Software Suite version 10.2, Dana-Farber Cancer Institute, Boston, MA) using k-nearest neighbor replacement and average linkage using Euclidean distance hierarchical clustering.

For ELISA, ICS, qPCR, proteinuria and kidney histology data, significant differences between groups were determined by Mann-Whitney test. For phosphoflow data, significant differences between groups were established as FDR <0.05, calculated by the Benjamini-Hochberg method for multiple hypothesis testing correction.